Alert sending rules

3 min read · Notifications

ePulz.io does not send an alert on every short fluctuation. The goal is to notify you only about real problems and not to overwhelm you with false alarms. In this article we explain when an alert is sent, how repeated attempts work, the recovery e-mail (recovery confirmation) and Quiet hours.

When an alert is triggered

An alert is sent when the monitor detects a problem - for example:

The server responds with a different HTTP code than expected (200 by default, which means OK).
The response takes longer than the configured time limit.
The response is missing a keyword that you set in the HTTP monitor (the Expected keyword field).
The SSL certificate is about to expire (warnings are sent 30, 14, 7, 3 and 1 day in advance).
The domain registration is about to expire (warnings are sent 30, 14, 7, 3 and 1 day in advance).
A Heartbeat monitor (waiting for a regular "heartbeat" from your job) did not receive the expected ping even within the grace period.

Protection against false alarms (anti-flap)

So that you do not get an alert on every short network "twitch", you can set a Number of failures before alerting value in each monitor:

1 - an immediate alert right on the first failure (the most sensitive).
2 or 3 - the recommended setting. It filters out short network interruptions without letting a real outage slip through.

If your server occasionally has brief fluctuations, we recommend a value of 2 or 3 as the default choice.

Verification from multiple regions

If more probes are available, we verify an outage from multiple regions and send the alert only after it is confirmed by at least two regions. This filters out local network problems (for example an outage of a single connectivity provider in another country) that do not actually affect you.

Recovery alert

When the monitor responds correctly again after an outage, we send you a recovery alert with the total duration of the outage. This gives you an immediate picture of whether you need to act or whether everything is already resolved. You can turn off recovery e-mails in the settings if an alert at the time of the outage is enough for you.

No duplicate alerts

During a single outage you receive only one alert. Even if the monitor fails several times in a row, you will not get dozens of e-mails. We automatically merge duplicate alerts (the same outage across different channels). We send a new alert only on the next separate outage, that is after the monitor has restored the service in the meantime.

Quiet hours

During quiet hours we do not send alerts - they will not arrive even retroactively after they end. For critical alerts use a webhook, which bypasses quiet hours. You set quiet hours and notification channels in the Notifications section:

Recommendation: For most websites the ideal setting is 2 failures before alerting and recovery alerts enabled. For critical infrastructure choose 1 and make use of verification from multiple regions.

Was this guide helpful?